# syntax=docker/dockerfile:1
FROM nvidia/cuda:13.1.0-devel-ubuntu26.04 AS base

RUN apt-get update && apt-get install -y --no-install-recommends \
    python3 python3-pip python3-dev python3-venv \
    gcc g++ wget curl jq \
    && rm -rf /var/lib/apt/lists/*

RUN ln -sf /usr/bin/python3 /usr/bin/python

# vLLM with NVFP4/modelopt support (Blackwell SM120) + flashinfer for the FP4 kernels
RUN pip3 install --no-cache-dir vllm flashinfer-python

RUN python3 -c "import vllm; print('vLLM', vllm.__version__)"

FROM base AS runtime
ENV CC=/usr/bin/gcc
ENV TORCH_MATMUL_PRECISION=high
ENV VLLM_USE_FLASHINFER_SAMPLER=1
EXPOSE 8000

COPY entrypoint.sh /entrypoint.sh
RUN chmod +x /entrypoint.sh
ENTRYPOINT ["/entrypoint.sh"]
CMD []